Why I Built the Senate Trade Tracker
The STOCK Act — Stop Trading on Congressional Knowledge Act — requires members of Congress to publicly disclose personal stock trades within 45 days of execution. In theory, this means the public can see whether lawmakers are trading in industries they regulate. In practice, these disclosures are buried on a government website, filed as PDFs, and largely ignored.
That felt like a problem worth solving.
What the Senate Trade Tracker does
The Senate Trade Tracker is a full-stack web app that monitors STOCK Act filings automatically. A data pipeline runs daily, fetching new disclosures from the Senate’s public database. When a new filing is detected, the system extracts the trade details — senator name, ticker, transaction type, amount range, and date — and passes them to the Claude AI API for analysis.
The AI generates a plain-English summary of each trade: what sector the company operates in, whether the senator sits on any relevant committees, and any notable context about the timing of the trade relative to public news. The summary is then delivered to a Telegram channel that anyone can follow.
The app also maintains a searchable database of all trades going back to the project’s launch, so you can filter by senator, ticker, date range, or transaction type.
Why this matters
The STOCK Act passed in 2012 with bipartisan support and genuine popular enthusiasm. The idea was simple: if a senator is on the Armed Services Committee and buys defense stocks right before a major contract announcement, the public should know. Disclosure is the mechanism Congress chose to enforce that accountability.
But disclosure only works if people can actually find and interpret the filings. A PDF buried in a government database, filed weeks after the trade occurred, isn’t very useful. Making those disclosures machine-readable, searchable, and summarized in plain language is a meaningful improvement — and it’s exactly the kind of work that AI is well suited for.
The technical stack
The frontend is built with React and TypeScript, hosted on Vercel at trades.errabot.com. The backend uses Express and Supabase for data storage. The AI analysis is generated via the Anthropic Claude API. A GitHub Actions workflow runs the data pipeline on a daily schedule, checking for new filings and processing them automatically.
All code is written and maintained by one person. The project has no funding, no team, and no editorial agenda — just a genuine interest in making public information more accessible.
How to follow along
You can browse the full trade database at trades.errabot.com. New AI-analyzed summaries are posted to the project’s Telegram channel whenever a new Senate filing is detected. The goal is to make congressional financial transparency as frictionless as possible — so that the accountability the STOCK Act promises is actually achievable in practice.
For more information on StOCK ACT Disclosures: https://www.errabot.com/what-senators-are-actually-trading-patterns-in-stock-act-disclosures/